Closed pkayokay closed 6 months ago
Excellent! Thanks!
@excid3 I found out we could alternatively update the timeout as well in the config. If the solution here is inconsistent we can also up the timeout time. Adding this as a comment here in case we need it in the future.
{
concurrency: 1,
files: "test/**/*_test.js",
nodeResolve: true,
filterBrowserLogs,
mimeTypes: {},
plugins: [],
testFramework: {
config: {
timeout: '10000',
},
},
}
Cool, it seems pretty stable now with concurrency of 1 and we only have a few tests so it's fast enough.
I don't really understand the flakiness having more concurrency though. 🤷♂️
One thing I did notice yesterday, one test was consistently failing because we initialize the fixture before Stimulus in the test. Swapping the order to initialize the Stimulus component and then load the fixture made it reliable. Maybe that was the source of it?
This may help reduce the failures we're seeing in CI and local. I actually got pretty consistent results of passing tests locally by reducing the concurrency option to 1.
Similar issue found here